When the dust settled on the 2024 US presidential election, prediction market Polymarket saw a surprising surge in trading volume—over $600 million. Yet, much of this activity revolved around long-shot candidates who ultimately lost.

In Ethiopia, the trading pattern was even stranger. The winning party’s leader, Abiy Ahmed, saw minimal betting, while his chief of staff, Shimelis Abdisa, attracted millions. This raised eyebrows among experts, sparking debate over whether these markets truly reflect public opinion.

“Nearly all of the trading volume—about $284 million—was from pairs of accounts repeatedly buying and selling with each other,” explains Chris Park, a former trader now analyzing such patterns. This high volume of seemingly pointless activity has led to suspicions of manipulation, challenging the notion that prediction markets are ‘truth machines’.

The peculiar trading dynamics highlight the complexity of these markets. While they can offer valuable insights, the 2024 and 2023 elections suggest they might also be susceptible to manipulation and speculative betting.

As experts grapple with these findings, the question remains: can prediction markets ever truly predict the future?